XPost: misc.consumers   
   From: fritz@spamexpire-202511.rodent.frell.theremailer.net   
      
   BBC probe finds AI chatbots mangle nearly half of news summaries   
      
   Google Gemini worst offender with 76% error rate   
      
   A major study [PDF] led by the BBC on behalf of the European   
   Broadcasting Union (EBU) found that OpenAI's ChatGPT, Microsoft   
   Copilot, Google Gemini, and Perplexity misrepresented news content in   
   almost half of the cases.   
      
   An analysis of more than 3,000 responses from the AI assistants found   
   that 45 percent of answers given contained at least one significant   
   issue, 31 percent had serious sourcing problems, and a fifth had "major   
   accuracy issues, including hallucinated details and outdated   
   information."   
      
   When accounting for smaller slip-ups, a whopping 81 percent of   
   responses included a mistake of some sort.   
      
   Gemini was identified as the worst performer, with researchers   
   identifying "significant issues" in 76 percent of responses it provided   
   – double the error rate of the other AI bots.   
      
   The researchers blamed this on Gemini's poor performance in sourcing   
   information, with researchers finding significant inaccuracies in 72   
   percent of responses. This was three times as many as ChatGPT (24   
   percent), followed by Perplexity and Copilot (both 15 percent).   
      
   More here:   
   https://www.theregister.com/2025/10/24/bbc_probe_ai_news/   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)